Blog: Elo Hell - Fact or Fiction?
posted by Gaegamel, 1 year agoElo Hell. Does it or does it not exist? Let's take a closer look...
We've all been there. You're minding your own business playing a ranked game, when suddenly everything goes wrong and you simply end up losing the game and elo respectively. Every single one of us has, and there is a particlular elo range where every game seems to have similar features and events from one game to the next. This is what a major part of the community consider as "elo hell", and subsequently manifest their frustration about this topic inside of forum-threads.
But how, or when, does this ideaology come into play? And does elo hell actually exist? Or is it just the shared opinion of objectively seen poor gamers?
Well, let's take a look at ELO. Apart from League of Legends, ELO is used as a ranking system in many other competitive activities, such as soccer (football), chess, tennis, etc. Exclusively in the examples given, the elo is calculated differently. Nevertheless, it shows the current estimated skill-level of the competitor(s) in comparison to others who are performing in the same activity.
Having a little bit of background information in mind, let's take a closer look at the actual mechanic. Whenever you lose a game, you lose elo. Just as you lose the elo your victorious opponents gain it. This also obviously occurs the other way round. This also occurs not only in ranked, but in normal games alike. Rumors concerning normal game elo have been clarified and normal elo does exist, although this information is not visible to the gamer themself. There are even programs out there used to determine your normal game elo. In general, LoL's elo is determined by mathematical formulae of which it's full dependencies are unknown to us.
The most important part (or "side-effect") of the game is this so-called "elo hell". After a few brief searches of the official LoL forums using the keyword "elo hell" you will find numerous threads created by players complaining about their elo and how they managed to fall to that level. Most of the players reside in the elo range of 900 to 1200. Having this in mind, you could assume that elo hell exists and that it ranges from 860 to 1250 elo.
This idea can be clarified using two arguments: The quantity and the quality of the LoL-gaming community (the players participating in ranked games).
Let's take a look at the "elo hell" range of 860-1200. Once a usual player hits level 30, they start with a rating named "not ranked“. This doesn't mean they have a ranking of 0 or anything alike, but 1250. Besides, your elo gain is increased in the first ten games. This means that you gain or lose more than you would usually do (after those first ten games).
In effect, losing a game at the beginning is a hard impact on your elo. Still, this does not really explain the situation properly. Thus, let's take a look at the requirements of playing ranked.
You're allowed to play ranked games upon reaching summoner level 30 and by owning 16 champions. Taking the average experience gain, you usually need to play 150-250 games (not wins) as a minimum to reach level 30 and to have unlocked around that number of champions, if not, more. Once you've done that, you start at the same elo as another player who may have potentially played something in the region of 1000 or even 2000 normal games. This leads to a sinkhole of skill levels in a certain elo area, ranging from generally unskilled players to great players.
So a potential conclusion to the scenario is that pure luck decides the first couple of your games and their outcomes.
Therefore the thought arises that Riot have made a bad job in designing this elo-system. However Riot has taken advantage of an old mathematical law, the Law of Large Numbers (LLN).
The LLN is based on averages and simple mathematics. Taking a die as an example, you have 6 possible sides to land on, hence showing 6 different values. If you consider the probability of all 6 values, you will notice that each one of them has a 16.666% potential of appearing after the previous roll.
This of course is theoretical probability, which means that it would probably not be accurate in a small amount of rolls. However, aside from other physical and/or situational factors, you can expect your theoretical probability to be more accurate after a higher number of rolls.
How can this be applied to LoL?
Instead of dice, we take games and apply the LLN. For instance, let's assume that everyone has the same skill level, except that one person who complains – they have a higher one. Considering those factors, he may or may not lose his first games, But in the end, the LLN applies accordingly and their elo would adjust over time (as the amount of games they have under their belt increases) played based on their skill level, and theoretically bringing them out of "elo hell".
To conclude: Objectively seen, elo hell does not exist - or at least not as how people are describing it: a big black hole, sucking everyone in and leaving no way of escape. You can, on the contrary, look at it as a vacuum of skill levels, which requires patience in order to escape from it. Applying this understanding once more would show that once you do, you're out - assuming you have the necessary skill level. Thus, elo hell is not a phenomenom in a general area, but an individual aspect of every LoL-players' gaming life.
Sources: Wikipedia: Elo rating, LoL-rating tool, Wikipedia: Law of large numbers, Preview Picture Source, Elo-hell picture source
- Escape your 'Elo Hell'! Part 1
- Season 2 Rewards, ELO Changes & Badges
- Monster Gaming: ELO Boosting Scandal
- Azubu TPA Lilballz Banned For Elo Boosting
- Elo-Boosters in the LCS Receive FINAL WARNING from Riot!
OMG vs. LGD tomorrow PE vs. eP tomorrow Dig vs. Crs 20h 46m CLG vs. CST 18h 46m WE vs. LGD 10h 46m SG Blue vs. CJE Frost 9h 46m RC.HZ vs. iG 7h 46m SK vs. Fnatic ROCCAT vs. Alliance GG vs. Wolves SK vs. ROCCAT Fnatic vs. Alliance SG Ozone vs. SKT T1 K ROCCAT vs. GG Alliance vs. Wolves LMQ vs. C9.Tempest Crs.Ac vs. coL.Black OMG vs. eP PE vs. EG LMQ vs. Crs.Ac